Lectures teach in fact lists. Step 1 tests in vignettes. This is the bridge.

A USMLE-style clinical vignette is the long-form item shape the boards use: a patient demographic, the presenting complaint, a focused history and exam, a labs or imaging block, then the actual question. Lectures teach the underlying fact (the enzyme deficiency, the receptor pharmacology, the loop-of-Henle physiology). The vignette wraps the same fact in a clinical scenario and asks you to recover the fact from the wrapping. Drilling vignettes from your lectures means uploading your professor's slide deck and having Studyly generate vignette-shaped questions whose answer keys are anchored to specific slides in your own deck, instead of practicing on UWorld items written from a generic blueprint.

Use them. UWorld and AMBOSS have the strongest item editors in the field, and on a real Step 1 timer their items are the closest practice to the exam. The reason to also drill vignettes from your own lecture deck is coverage: UWorld is written against the published Step 1 content outline, not against what your professor decided to emphasize this block. Your block exam, your shelf prep, and the half-mark questions on your real Step 1 that come from a topic UWorld treats lightly all reward drilling on your professor's actual material. Lecture-grounded vignettes close that coverage gap; UWorld closes the writing-quality gap. Use both.

That is exactly the cognitive transformation Step 1 tests. The slide says 'thick ascending limb of Henle is impermeable to water and actively reabsorbs NaCl'. The vignette says 'a 38-year-old presents with hyperosmolar urine; which segment of the nephron concentrates the medullary interstitium by being impermeable to water while actively reabsorbing solute?' The fact is the same. The wrapper is a clinical scenario plus a hidden cue (hyperosmolar urine implies the concentrating mechanism, which implies the medullary countercurrent multiplier, which implies the thick ascending limb). The case-style generator is built to produce that wrapping; the underlying topic-pin stays anchored to the slide.

Demographic (age plus relevant context like 'previously healthy' or '3-day post-op'), presentation (chief complaint plus duration), exam (the 1-2 findings the case turns on), labs or imaging (1-3 numbers or a one-line read), and question stem (most likely diagnosis, best next step, mechanism, or the underlying-physiology question). Real NBME items follow the same five-element shape, with the optional sixth element being a brief past medical history. Stems run roughly 70-150 words, which is what the generator targets to match the actual NBME item-length distribution.

Studyly tracks each card as a topic-pin, not as a frozen item. On revisit, an LLM pass regenerates the surface form against the same pin: a different demographic, a different presenting symptom that maps to the same underlying mechanism, a rotated distractor pool, the right-answer index moves. The internal pin stays fixed, so the spaced-repetition scheduler keeps tracking your retention of the underlying fact. By revisit five you have answered the same fact wrapped in five different clinical scenarios. If you actually learned the slide, you still get it right; if you were memorizing the wording of revisit one, this is where the wheels come off.

A quality gate runs at revisit time, the same four-criterion rubric that scores the initial generation: factual correctness against the source slide, stem clarity, distractor plausibility, and question-type coverage. A revisit that drops below the gate gets rolled back to the prior wording. The same rubric is what produced the 81.3 score on the held-out three-document eval, vs Unattle 78.0, Gauntlet 68.0, and Turbolearn 57.8. Methodology is on the quality page.

About 60 seconds for a typical 90-slide deck. The four generators (MCQ, free-response, case-style, image-occlusion on labeled figures) run in parallel, so the case-style track produces roughly 60-80 vignette-format cards from that deck. The total card count across all four formats is around 200, which sounds like a lot until you remember that each card is a 5-minute drill, not a 30-minute study session. Five minutes a night, one deck per day, is the default loop.

Both, with the leverage shifted. Step 1 vignettes are usually one-fact-per-stem with a basic-science answer; the wrapping is shorter and the cognitive jump from slide-fact to vignette is bigger. Step 2 CK vignettes are clinical-management questions with longer histories and exam blocks; the cognitive jump from slide-fact to vignette is smaller because the slide is already clinical. The case-style generator targets a longer stem on CK material when the source material is a clinical lecture (internal medicine block, pediatrics shelf prep, OB-GYN clerkship deck).

The unique value here assumes you have a source document. If you are studying from First Aid plus UWorld plus a video series with no professor's slides, the case-style generator still works on the source material you do have: a First Aid chapter, a typed lecture transcript, your own typed UWorld notes, a textbook chapter. The vignette template does not care whether the source is a slide deck or a chapter; it cares that there is a testable fact to wrap. The shorter the source, the fewer vignettes per upload, but the loop is the same.

No. The topic-pin defines the right answer; the regeneration pass changes only the wrapping (demographic, presentation, exam, lab values, distractor pool, right-answer index position). The fact under test does not move, which is the entire point of pinning the underlying concept and rephrasing the surface. If a regenerated stem somehow changed the answer, the rubric gate at revisit time would catch the factual-correctness drop and roll back to the prior wording.